Revision 35145
Added by Argiro Kokogiannaki over 9 years ago
| claim.php | ||
|---|---|---|
| 672 | 672 |
// return a result (object) containing publications and total |
| 673 | 673 |
private function _searchDOI($doi, $page, $size) {
|
| 674 | 674 |
try {
|
| 675 |
$request = self :: CROSSREF_API_DOI . urlencode( trim($doi)); |
|
| 676 |
JLog :: add('Requesting ' . $request, JLog :: INFO, self :: LOG);
|
|
| 677 |
$time = microtime(TRUE); |
|
| 678 |
$response = $this -> http -> get($request); |
|
| 679 |
JLog :: add('Received response in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 680 |
if ($response == NULL) |
|
| 681 |
throw new Exception('no HTTP response');
|
|
| 682 |
if ($response -> code != self :: HTTP_OK) |
|
| 683 |
throw new Exception('HTTP response code ' . $response -> code);
|
|
| 684 |
$res = array(); |
|
| 685 |
$res = json_decode(str_replace("-", "_", $response -> body));
|
|
| 686 |
$result = new JObject(); |
|
| 687 |
$result -> publications = array(); |
|
| 688 |
$result -> totalPublications = 0; |
|
| 689 |
$result -> totalDatasets = 0; |
|
| 690 |
if(isset($res -> status) && $res -> status === "ok" && isset($res -> message)){
|
|
| 691 |
|
|
| 692 |
$publication = $this -> _parseJsonDOI($res -> message); |
|
| 693 |
if($publication!=null){
|
|
| 694 |
$result -> publications[] = $publication; |
|
| 695 |
$result -> totalPublications = 1; |
|
| 696 |
} |
|
| 697 |
} |
|
| 675 |
$result = new JObject(); |
|
| 676 |
$result -> publications = array(); |
|
| 677 |
$result -> totalPublications = 0; |
|
| 678 |
$result -> totalDatasets = 0; |
|
| 679 |
$dois = explode(" ", preg_replace('/\s+/', ' ',$doi));
|
|
| 680 |
$unique_dois =array_slice( array_unique($dois),0,10); |
|
| 681 |
$pattern1 ='#\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&\'<>])\S)+)\b#';
|
|
| 682 |
$pattern2 ='#\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&\'<>])[[:graph:]])+)\b#';
|
|
| 683 |
if (preg_match($pattern1,$unique_dois[0])||preg_match($pattern2,$unique_dois[0])){
|
|
| 684 |
if(count($dois)>10){
|
|
| 685 |
$result = new JObject(); |
|
| 686 |
$result->publications = array(); |
|
| 687 |
$result->totalPublications = '-'; |
|
| 688 |
$result->totalDatasets = 0; |
|
| 689 |
$result->maxDoisExceeded = true; |
|
| 690 |
return $result; |
|
| 691 |
} |
|
| 692 |
JLog :: add('It;s a DOI!!!!!!!!! '.$dois[0] , JLog :: INFO, self :: LOG);
|
|
| 693 |
JLog :: add('Searching for DOIs ' , JLog :: INFO, self :: LOG);
|
|
| 694 |
foreach($unique_dois as $doi_){
|
|
| 695 |
JLog :: add('DOI::: '.$doi_ , JLog :: INFO, self :: LOG);
|
|
| 696 |
$publication = $this->_searchSingleDOI($doi_); |
|
| 697 |
if ($publication != null) {
|
|
| 698 |
$result->publications[] = $publication; |
|
| 699 |
} |
|
| 700 |
} |
|
| 701 |
}else{
|
|
| 702 |
JLog :: add('It;s NOOOOT a DOI!!!!!!!!! '.$dois[0] , JLog :: INFO, self :: LOG);
|
|
| 703 |
} |
|
| 704 |
|
|
| 705 |
$result -> totalPublications = count( $result -> publications); |
|
| 698 | 706 |
if($result -> totalPublications ===0){
|
| 699 | 707 |
//.'&rows='.$size.'&offset='.$page |
| 700 | 708 |
$request = "http://api.crossref.org/works?query=" . urlencode( trim($doi))."&rows=".$size."&offset=".($page-1)*$size; |
| ... | ... | |
| 732 | 740 |
return $result; |
| 733 | 741 |
} |
| 734 | 742 |
} |
| 743 |
public function searchSingleDOI($doi) {
|
|
| 744 |
if ($this -> cache -> getCaching()) {
|
|
| 745 |
$cacheId = self :: SEARCH_DOI_CACHE_ID . '.' . $doi; |
|
| 746 |
$results = $this -> cache -> get($cacheId, self :: CACHE_GROUP); |
|
| 747 |
if ($results === FALSE) {
|
|
| 748 |
$results = $this -> _searchSingleDOI($doi); |
|
| 749 |
if ($results !== NULL) |
|
| 750 |
$this -> cache -> store($results, $cacheId, self :: CACHE_GROUP); |
|
| 751 |
} |
|
| 752 |
} else |
|
| 753 |
$results = $this -> _searchSingleDOI($doi); |
|
| 754 |
return $results; |
|
| 755 |
} |
|
| 756 |
private function _searchSingleDOI($doi) {
|
|
| 757 |
try {
|
|
| 758 |
$request = self :: CROSSREF_API_DOI . urlencode( trim($doi)); |
|
| 759 |
JLog :: add('Requesting ' . $request, JLog :: INFO, self :: LOG);
|
|
| 760 |
$time = microtime(TRUE); |
|
| 761 |
$response = $this -> http -> get($request); |
|
| 762 |
JLog :: add('Received response in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 763 |
if ($response == NULL) |
|
| 764 |
throw new Exception('no HTTP response');
|
|
| 765 |
if ($response -> code != self :: HTTP_OK) |
|
| 766 |
throw new Exception('HTTP response code ' . $response -> code);
|
|
| 767 |
$res = array(); |
|
| 768 |
$res = json_decode(str_replace("-", "_", $response -> body));
|
|
| 769 |
if(isset($res -> status) && $res -> status === "ok" && isset($res -> message)){
|
|
| 770 |
$publication = $this -> _parseJsonDOI($res -> message); |
|
| 771 |
return $publication; |
|
| 772 |
} |
|
| 773 |
|
|
| 774 |
} catch (Exception $e) {
|
|
| 775 |
JLog :: add('Error performing DOI search (doi: ' . $doi . '): ' . $e -> getMessage(), JLog :: ERROR, self :: LOG);
|
|
| 776 |
$result = new JObject(); |
|
| 777 |
$result -> publications = array(); |
|
| 778 |
$result -> totalPublications = 0; |
|
| 779 |
$result -> totalDatasets = 0; |
|
| 780 |
JLog :: add('error parsing DOI record', JLog :: INFO, self :: LOG);
|
|
| 781 |
return null; |
|
| 782 |
} |
|
| 783 |
} |
|
| 735 | 784 |
private function _parseJsonDOI($message) {
|
| 736 | 785 |
$publication = new JObject(); |
| 737 | 786 |
$publication -> id = isset($message -> DOI)?$message -> DOI:NULL; |
| ... | ... | |
| 765 | 814 |
return NULL; |
| 766 | 815 |
|
| 767 | 816 |
} |
| 768 |
private function _searchDOIOld($doi, $page, $size) {
|
|
| 769 |
try {
|
|
| 770 |
$request = self :: CROSSREF_URL . urlencode(self :: DOI_SCHEME . trim($doi)); |
|
| 771 |
JLog :: add('Requesting ' . $request, JLog :: INFO, self :: LOG);
|
|
| 772 |
$time = microtime(TRUE); |
|
| 773 |
$response = $this -> http -> get($request); |
|
| 774 |
JLog :: add('Received response in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 775 |
if ($response == NULL) |
|
| 776 |
throw new Exception('no HTTP response');
|
|
| 777 |
if ($response -> code != self :: HTTP_OK) |
|
| 778 |
throw new Exception('HTTP response code ' . $response -> code);
|
|
| 779 |
$document = new DOMDocument(); |
|
| 780 |
$document -> recover = TRUE; |
|
| 781 |
if ($document -> loadXML(trim($response -> body)) == FALSE) |
|
| 782 |
throw new Exception('invalid XML response');
|
|
| 783 |
$xpath = new DOMXPath($document); |
|
| 784 |
if ((($crossrefNodes = $xpath -> query('/doi_records/doi_record/crossref')) == FALSE) || (($crossrefNode = $crossrefNodes -> item(0)) == NULL))
|
|
| 785 |
throw new Exception('error parsing DOI record');
|
|
| 786 |
if (($journalNodes = $xpath -> query('./journal', $crossrefNode)) == FALSE)
|
|
| 787 |
throw new Exception('error parsing DOI record');
|
|
| 788 |
if (($bookNodes = $xpath -> query('./book', $crossrefNode)) == FALSE)
|
|
| 789 |
throw new Exception('error parsing DOI record');
|
|
| 790 |
if (($conferenceNodes = $xpath -> query('./conference', $crossrefNode)) == FALSE)
|
|
| 791 |
throw new Exception('error parsing DOI record');
|
|
| 792 |
/* if (($dissertationNodes = $xpath -> query('./dissertation', $crossrefNode)) == FALSE)
|
|
| 793 |
throw new Exception('error parsing DOI record');
|
|
| 794 |
if (($reportPaperNodes = $xpath -> query('./report-paper', $crossrefNode)) == FALSE)
|
|
| 795 |
throw new Exception('error parsing DOI record');
|
|
| 796 |
if (($standardNodes = $xpath -> query('./standard', $crossrefNode)) == FALSE)
|
|
| 797 |
throw new Exception('error parsing DOI record');
|
|
| 798 |
if (($saComponentNodes = $xpath -> query('./sa_component', $crossrefNode)) == FALSE)
|
|
| 799 |
throw new Exception('error parsing DOI record');
|
|
| 800 |
if (($databaseNodes = $xpath -> query('./database', $crossrefNode)) == FALSE)
|
|
| 801 |
throw new Exception('error parsing DOI record');*/
|
|
| 802 |
if (($errorNodes = $xpath -> query('./error', $crossrefNode)) == FALSE){
|
|
| 803 |
//throw new Exception('error parsing DOI record');
|
|
| 804 |
$result = new JObject(); |
|
| 805 |
$result -> publications = array(); |
|
| 806 |
$result -> totalPublications = 0; |
|
| 807 |
$result -> totalDatasets = 0; |
|
| 808 |
JLog :: add('error parsing DOI record', JLog :: INFO, self :: LOG);
|
|
| 809 |
return $result; |
|
| 810 |
} |
|
| 811 |
if (($journalNode = $journalNodes -> item(0)) != NULL) {
|
|
| 812 |
$result = $this -> parseJournal($doi, $xpath, $journalNode); |
|
| 813 |
JLog :: add('Retrieved journal in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 814 |
$result -> totalPublications = 1; |
|
| 815 |
$result -> totalDatasets = 0; |
|
| 816 |
return $result; |
|
| 817 |
} else if (($bookNode = $bookNodes -> item(0)) != NULL) {
|
|
| 818 |
$result = $this -> parseBook($doi, $xpath, $bookNode); |
|
| 819 |
JLog :: add('Retrieved book in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 820 |
$result -> totalPublications = 1; |
|
| 821 |
$result -> totalDatasets = 0; |
|
| 822 |
return $result; |
|
| 823 |
}else if (($conferenceNode = $conferenceNodes -> item(0)) != NULL){
|
|
| 824 |
$result = $this -> parseConference($doi, $xpath, $conferenceNode); |
|
| 825 |
JLog :: add('Retrieved conference in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 826 |
$result -> totalPublications = 1; |
|
| 827 |
$result -> totalDatasets = 0; |
|
| 828 |
return $result; |
|
| 829 |
/* else if (($dissertationNode = $dissertationNodes -> item(0)) != NULL) |
|
| 830 |
return dissertation; |
|
| 831 |
else if (($reportPaperNode = $reportPaperNodes -> item(0)) != NULL) |
|
| 832 |
return reportPaper; |
|
| 833 |
else if (($standardNode = $standardNodes -> item(0)) != NULL) |
|
| 834 |
return standard; |
|
| 835 |
else if (($saComponentNode = $saComponentNodes -> item(0)) != NULL) |
|
| 836 |
return saComponent; |
|
| 837 |
else if (($databaseNode = $databaseNodes -> item(0)) != NULL) |
|
| 838 |
return database;*/ |
|
| 839 |
} else if (($errorNode = $errorNodes -> item(0)) != NULL) {
|
|
| 840 |
$result = new JObject(); |
|
| 841 |
$result -> publications = array(); |
|
| 842 |
$result -> totalPublications = 0; |
|
| 843 |
$result -> totalDatasets = 0; |
|
| 844 |
JLog :: add('Retrieved 0 DOI records in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
|
| 845 |
return $result; |
|
| 846 |
} else{
|
|
| 847 |
throw new Exception('error parsing DOI record');
|
|
| 848 |
|
|
| 849 |
|
|
| 850 |
} |
|
| 851 |
} catch (Exception $e) {
|
|
| 852 |
JLog :: add('Error performing DOI search (doi: ' . $doi . '): ' . $e -> getMessage(), JLog :: ERROR, self :: LOG);
|
|
| 853 |
$result = new JObject(); |
|
| 854 |
$result -> publications = array(); |
|
| 855 |
$result -> totalPublications = 0; |
|
| 856 |
$result -> totalDatasets = 0; |
|
| 857 |
JLog :: add('error parsing DOI record', JLog :: INFO, self :: LOG);
|
|
| 858 |
return $result; |
|
| 859 |
} |
|
| 860 |
} |
|
| 861 |
|
|
| 817 |
|
|
| 862 | 818 |
// Search for a dataset DOI using caching if enabled. |
| 863 | 819 |
// $doi the dataset DOI to search for |
| 864 | 820 |
// return a result (object) containing datasets and total |
| 865 |
/*private function _searchDataCite($doi) {
|
|
| 821 |
private function _searchDataCiteWithDoi($doi) {
|
|
| 866 | 822 |
try {
|
| 867 | 823 |
$request = self :: DATACITE_URL . urlencode(trim($doi)); |
| 868 | 824 |
JLog :: add('Requesting ' . $request, JLog :: INFO, self :: LOG);
|
| ... | ... | |
| 871 | 827 |
JLog :: add('Received response in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
| 872 | 828 |
if ($response== NULL) |
| 873 | 829 |
throw new Exception('no HTTP response');
|
| 874 |
if ($response -> code == self :: HTTP_NOT_FOUND) { // no dataset found; just return empty result
|
|
| 830 |
/*if ($response -> code == self :: HTTP_NOT_FOUND) { // no dataset found; just return empty result
|
|
| 875 | 831 |
$result = new JObject(); |
| 876 | 832 |
$result -> datasets = array(); |
| 877 | 833 |
$result -> totalDatasets = 0; |
| 878 | 834 |
JLog :: add('Retrieved 0 DataCite records in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
| 879 | 835 |
return $result; |
| 880 |
} |
|
| 836 |
}*/
|
|
| 881 | 837 |
if ($response -> code != self :: HTTP_OK) |
| 882 | 838 |
throw new Exception('HTTP response code ' . $response -> code);
|
| 883 | 839 |
$document = new DOMDocument(); |
| ... | ... | |
| 920 | 876 |
if ($author -> fullName != NULL) |
| 921 | 877 |
$dataset -> authors[] = $author; |
| 922 | 878 |
} |
| 923 |
$result = new JObject(); |
|
| 879 |
/*$result = new JObject();
|
|
| 924 | 880 |
$result -> datasets = array(); |
| 925 | 881 |
$result -> totalDatasets = 0; |
| 926 | 882 |
if (($dataset -> id != NULL) || ($dataset -> url != NULL) || ($dataset -> title != NULL) || ($dataset -> authors != NULL) || ($dataset -> year != NULL)) {
|
| 927 | 883 |
$result -> datasets[] = $dataset; |
| 928 | 884 |
$result -> totalDatasets = 1; |
| 929 |
} |
|
| 885 |
}*/
|
|
| 930 | 886 |
JLog :: add('Retrieved DataCite record in ' . (microtime(TRUE) - $time) . ' s', JLog :: INFO, self :: LOG);
|
| 931 |
return $result;
|
|
| 887 |
return $dataset;
|
|
| 932 | 888 |
} catch (Exception $e) {
|
| 933 | 889 |
JLog :: add('Error performing DataCite search (doi: ' . $doi . '): ' . $e -> getMessage(), JLog :: ERROR, self :: LOG);
|
| 934 | 890 |
return NULL; |
| 935 | 891 |
} |
| 936 |
}*/
|
|
| 892 |
} |
|
| 937 | 893 |
private function _searchDataCite($doi, $page, $size) {
|
| 938 |
try {
|
|
| 894 |
$result = new JObject(); |
|
| 895 |
$result->datasets = array(); |
|
| 896 |
$result->totalPublications = 0; |
|
| 897 |
$result->totalDatasets = 0; |
|
| 898 |
try {
|
|
| 899 |
$dois = explode(" ", preg_replace('/\s+/', ' ',$doi));
|
|
| 900 |
|
|
| 901 |
$unique_dois =array_slice( array_unique($dois),0,10); |
|
| 902 |
$pattern1 = '#\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&\'<>])\S)+)\b#';
|
|
| 903 |
$pattern2 = '#\b(10[.][0-9]{4,}(?:[.][0-9]+)*/(?:(?!["&\'<>])[[:graph:]])+)\b#';
|
|
| 904 |
if (preg_match($pattern1, $unique_dois[0]) || preg_match($pattern2, $unique_dois[0])) {
|
|
| 905 |
if(count($dois)>10){
|
|
| 906 |
$result = new JObject(); |
|
| 907 |
$result->datasets = array(); |
|
| 908 |
$result->totalPublications = 0; |
|
| 909 |
$result->totalDatasets = '-'; |
|
| 910 |
$result->maxDoisExceeded = true; |
|
| 911 |
return $result; |
|
| 912 |
} |
|
| 913 |
JLog :: add('It;s a dataset DOI!!!!!!!!! ' . $dois[0], JLog :: INFO, self :: LOG);
|
|
| 914 |
JLog :: add('Searching for DOIs in Datacite', JLog :: INFO, self :: LOG);
|
|
| 915 |
foreach ($unique_dois as $doi_) {
|
|
| 916 |
JLog :: add('DOI::: ' . $doi_, JLog :: INFO, self :: LOG);
|
|
| 917 |
$dataset = $this->_searchDataCiteWithDoi($doi_); |
|
| 918 |
if ($dataset != null) {
|
|
| 919 |
$result->datasets[] = $dataset; |
|
| 920 |
} |
|
| 921 |
} |
|
| 922 |
}else{
|
|
| 923 |
|
|
| 924 |
JLog :: add('It;sNOOOOOOOT a dataset DOI!!!!!!!!! ' . $dois[0], JLog :: INFO, self :: LOG);
|
|
| 925 |
} |
|
| 926 |
$result->totalDatasets = count($result->datasets); |
|
| 927 |
if($result->totalDatasets!==0){
|
|
| 928 |
return $result; |
|
| 929 |
} |
|
| 939 | 930 |
$request = self :: DATACITE_URL_KEYWORD. urlencode(trim($doi)).self :: DATACITE_URL_KEYWORD_REST."&rows=".$size."&start=".($page-1)*$size; |
| 940 | 931 |
JLog :: add('Requesting ' . $request, JLog :: INFO, self :: LOG);
|
| 941 | 932 |
$time = microtime(TRUE); |
| ... | ... | |
| 962 | 953 |
throw new Exception('error parsing DataCite record');
|
| 963 | 954 |
$numFound=((($numFoundNode)) == NULL) ? NULL : trim($numFoundNode -> nodeValue); |
| 964 | 955 |
if ((($resultNodes = $xpath -> query('/response/result/doc')) == FALSE) || (($resultNode = $resultNodes -> item(0)) == NULL))
|
| 965 |
throw new Exception('error parsing DataCite record');
|
|
| 966 |
$result = new JObject(); |
|
| 967 |
$result -> datasets = array(); |
|
| 968 |
$result -> totalDatasets = 0; |
|
| 956 |
throw new Exception('error parsing DataCite record');
|
|
| 969 | 957 |
foreach ($resultNodes as $resultNode) {
|
| 970 | 958 |
if (($idNodes = $xpath -> query('./str[@name = "doi"]/text()', $resultNode)) == FALSE)
|
| 971 | 959 |
throw new Exception('error parsing DataCite record');
|
| ... | ... | |
| 1415 | 1403 |
$publication -> description = (($shortDescriptionNode = $shortDescriptionNodes -> item(0)) == NULL) ? NULL : trim($shortDescriptionNode -> nodeValue); |
| 1416 | 1404 |
if (($author -> lastName != NULL) || ($author -> firstName != NULL)) |
| 1417 | 1405 |
$publication -> authors[] = $author; |
| 1418 |
if ((($doiNode = $doiNodes -> item(0)) != NULL) && (($doiResult = $this -> searchDOI(trim($doiNode -> nodeValue),10,1)) != NULL) && ($doiResult -> totalPublications > 0)) { // resolve via DOI
|
|
| 1406 |
if ((($doiNode = $doiNodes -> item(0)) != NULL) && (($doiResult = $this -> searchSingleDOI(trim($doiNode -> nodeValue))) != NULL)) { // resolve via DOI
|
|
| 1419 | 1407 |
if ($publication -> url == NULL) // set URL if missing |
| 1420 |
$publication -> url = $doiResult -> publications[0] -> url;
|
|
| 1408 |
$publication -> url = $doiResult -> url; |
|
| 1421 | 1409 |
if ($publication -> title == NULL) // set title if missing |
| 1422 |
$publication -> title = $doiResult -> publications[0] -> title;
|
|
| 1410 |
$publication -> title = $doiResult -> title; |
|
| 1423 | 1411 |
|
| 1424 |
foreach ($doiResult -> publications[0] -> authors as $doiAuthor) { // merge authors
|
|
| 1412 |
foreach ($doiResult -> authors as $doiAuthor) { // merge authors
|
|
| 1425 | 1413 |
if ((count($publication -> authors) == 0) || ($doiAuthor -> id != $publication -> authors[0] -> id) || ($doiAuthor -> lastName != $publication -> authors[0] -> lastName) || ($doiAuthor -> firstName != $publication -> authors[0] -> firstName) || ($doiAuthor -> fullName != $publication -> authors[0] -> fullName)) |
| 1426 | 1414 |
$publication -> authors[] = $doiAuthor; |
| 1427 | 1415 |
} |
| 1428 | 1416 |
if ($publication -> year == NULL) // set year if missing |
| 1429 |
$publication -> year = $doiResult -> publications[0] -> year;
|
|
| 1417 |
$publication -> year = $doiResult -> year; |
|
| 1430 | 1418 |
} |
| 1431 | 1419 |
if (($publication -> id != NULL) || ($publication -> title != NULL) || ($publication -> authors != NULL) || ($publication -> year != NULL) || ($publication -> description != NULL)) |
| 1432 | 1420 |
$result -> publications[] = $publication; |
Also available in: Unified diff
claims search: search in openaire/crossref/datacite with list of dois, up to 10 dois for crossref and datacite , datasources query: add (datasourcecompatibilityid <> hostedBy)